The present invention relates to a method and a system for spectacles frame image processing and recording for creating a data-base of a spectacles frame by cutting out an image of only a spectacles frame with high accuracy so as to synthesize an image of a human face and a frame image.
In a recently developed method, a variety of spectacles frame images are recorded and selectively synthesized with an image of a face, which is input through a television (to be referred to as TV hereinafter) camera, of a person who wants to select a spectacles frame. The user selects a spectacles frame while monitoring a synthetic image. In order to select a spectacles frame in this way, images of various spectacles frames must be recorded as a data-base. Therefore, only data of a frame portion is extracted from an input spectacles frame image, and the frame portion is cut out from the spectacles frame image in accordance with the data.
Conventionally, a spectacles frame image to be recorded is input through a TV camera or the like, a necessary portion, e.g., a predetermined rectangular region in which the spectacles frame image is placed is cut out using a cursor while displaying the spectacles frame image on a TV monitor, and image data of this portion is recorded. In this case, since an input image signal has various characteristics, an image signal of a background is not always constant, or an image signal of a spectacles frame is not always constant. Even if the image signal of the background is constant, edge enhancement inherent in a TV signal, i.e., enhancement processing is performed to a boundary between the background and the spectacles frame. For this reason, it is difficult to cut out a necessary image portion from the input image with high accuracy, and an edge of a cutout image portion sometimes becomes unnatural
In addition, conventionally, in order to create a data-base of a spectacles frame image, image data in a region including a spectacles frame are all cut out and recorded. However, of the spectacles frame image, the amount of image data of a frame itself is very small, and most of the remaining portion is background image data. For this reason, frame data must be effectively cut out to reduce an data amount of a portion including the spectacles frame, thereby recording a large number of spectacles frame data in a recording medium.
Moreover, when a spectacles frame image is to be synthesized with an image of a human face, a boundary of a synthetic image tends to become unnatural. That is, an image signal is generally represented by an NTSA (National Television System Committee) signal. Therefore, if a difference between colors or brightnesses of both the images is large, jaggy is produced in the boundary therebetween. In addition, when color lenses are used with a spectacles frame, it is practically difficult to confirm whether the spectacles frame fits a user's face.
Therefore, a demand has arisen for a method and a system for spectacles frame image processing and recording capable of cutting out only a necessary image portion and synthesizing a cutout image and another image without unnaturalness.